Abstract
One of the most common types of fluid storage tanks are rectangular concrete tanks placed on the ground. Ensuring the seismic performance of these tanks is crucial as, besides damage and losses during a crisis, they can lead to other crises such as fires or toxic material leaks. One pivotal factor influencing the seismic behavior of these tanks is the height of the freeboard. A comprehensive understanding of how freeboard height impacts the impulsive and convective masses is essential for the effective design and dynamic analysis of rectangular tanks. While numerous numerical and experimental studies have explored this field, the influence of various filling levels, fluid properties, and tank geometries under seismic loading conditions requires further investigation. In this study, an experimental model placed on a shaking table and a numerical model developed in ansys software were subjected to various earthquake records. The effects under seismic loading were assessed across different fill levels, tank geometries, and freeboard heights. In contrast to previous studies, the new results revealed a nonlinear relationship between freeboard height and variations in impulsive and convective masses. Finally, the proposed machine learning model accurately developed this nonlinear relationship, which can be utilized to improve current design practices for rectangular tanks in seismic-prone areas.